Position switchers in Oakland

By Jason Wojciechowski on September 2, 2012 at 6:40 PM

So I have this thing, you know, a blog? You've heard of it? No, it really exists, I swear.

So friend o' the blog Albert (Albie) Lau emailed a while back (I won't tell you how long) asking some questions. I won't quote it here because I'm writing without having asked his permission to do so, but the basic gist of it were like so:

What do the A's gain or lose by playing Pennington at second instead of Weeks? (Subsidiary question: please explain BVORP.)

Thoughts on the defensive performance of A's who have switched positions and how to measure their effectiveness?

To the first, I think the A's are playing a delicate balancing game. They're trying to get into the playoffs, but they're not selling out the future, either. This is part of why you didn't see Dan Straily or A.J. Griffin traded for someone like Mike Aviles. Some portion of the question about Pennington and Weeks, then, is how much the A's are motivated by 2012 considerations and how much they're trying to get Weeks straightened out in a low-pressure, high-coaching environment so that he can be a contributor for the next four or five years at second base.

If we consider the move for 2012 only, here are some numbers:

Player TAv TAv (PECOTA ROS) RAA (PECOTA ROS) Pennington .232 .249 -1.4 Weeks .233 .243 -2.2 TAv is, of course, True Average, Baseball Prospectus's offensive value metric. It is designed to look like batting average and .260 is the league average mark. The third column is what PECOTA, BP's forecast system, figures each player will put up in TAv for the rest of the year. The last column takes that rest-of-season projection and converts it to runs above average (or below average in this case) by the simple conversion of subtracting .260 (average), dividing by 0.9, and multiplying by plate appearances. (I used 115, which is based on an assumption that one player would get all the time at the spot. In Pennington's case, this is of course not true, as he is dividing time with Adam Rosales, but it allows an apples-to-apples comparison.)

So! The expected offensive difference between the two is negligible (less than a run over the rest of the year), as is the difference between how each has performed in the majors thus far.

Defense is a big question, but it's not an easy one to answer. FRAA, BP's defensive metric, which, unlike UZR and DRS, is available on publicly available play-by-play data, says that Weeks has been eight runs below average at second. (For what it's worth, FRAA splits the difference between DRS's -14 and UZR's -3 on Weeks. No data, sliced no way, thinks he's been good at the position, in other words.)

Will Pennington be worse than Weeks would be defensively? It's hard to imagine how he couldn't be, in a certain sense. To the extent that he has certain tools advantages on Weeks, you can imagine them being wiped out by unfamiliarity with the position or flat-out uselessness (by which I mean that his arm, cannon though it may be, isn't quite as important given how much closer to first he'll be playing — that's not, of course, to say that it's entirely useless, as flat-footed turns on double plays or balls up the middle still require a good arm, but the percentage of plays in which the arm strength matters is greatly reduced).

On the other hand, Pennington seems to be a good defensive shortstop, and the typical position-switcher off of short does better at other positions, given that short is the toughest position to play well of the three infield positions (with first base requiring almost an entirely different skill-set than the other three, and thus being excluded as sui generis).

So I guess my answer is that it's hard to say. Pennington's looked fine in the few games he's played there since the Weeks demotion, but being completely honest with you, I'd have said that Weeks looked fine, too. It was noticeable that he was not getting to balls that we all remember Mark Ellis converting into outs, but I could easily chalk that up to dropping from a superlative defensive player to a merely average one. I would not, with my eyes, have labeled Weeks below average. (Which is not me attacking the stats! Far from it, in fact: it's actually me saying that I'm a horrifyingly bad scout.)

As for BVORP, the actual derivation would require some time, but I think I can summarize in a hopefully useful way.

First, the "B" stands for "Batter," so we know we're talking only about position players here. (Or, more accurately, for players acting as position players. Pitchers accumulate BVORP, too, when they hit.)

Second, here is a list of components of BVORP:

Batting value, derived from TAv (which includes a park adjustment) Baserunning (not just steals and caught, but other advancement as well) Position adjustment Playing time / replacement level The third bullet, the position adjustment, is why comparing Pennington's BVORP to Weeks's in attempting to answer the question of how the position switch will play out is not something I'd favor. Pennington's BVORP is calculated based on him playing shortstop. Weeks's is calculated based on him playing second. Thus, the baselines are different. What BVORP is getting at is overall player value (before considering fielding).

If we're just dropping in one player for another, then, it makes more sense to simply look at the two players' hitting and running compared to each other, and then do what we can to guess about the defense.

To the second question, I'm OK using FRAA with an understanding of the size of the error bars around the figure printed on the Baseball Prospectus site. (See Colin Wyers here, for instance, on the MOE.) The two main guys who switched positions this year because of injuries and underperformance elsewhere on the roster have been Brandon Moss (-2 FRAA, which includes a little bit of time in the outfield) and Josh Donaldson (+4 FRAA).

Now, we're talking about guys with tiny amounts of time in the field. With as few balls in play as they've seen, the margins of error are rather large. That said, I can't exactly disagree with either assessment: Moss still seems to be learning which balls he can go after and which he can't, and Donaldson appears surprisingly spry for a catcher, with plenty of arm.

Yoenis Cespedes has also moved positions mid-year, but it's harder to say with him because FRAA is not available broken out by position — the figure you see on his player card at BP is a sum for both center and left. I would say that he seems to have made fewer gruesome reads (and thus taken fewer gruesome routes) since moving to left, but I'm not convinced that this means left is his better position long-term — in theory, he just needed an adjustment to the way balls come off of bats in the American major leagues, and he was in the process of making such adjustment when he made the move. On this hypothesis, he can move back to center whenever the A's want him to and make a seamless transition. (Though I'd strike a note of caution: the A's have won nine games in a row and are in a very good position for the playoffs. This has a way of making me giddy about the team in general, which trickles down to my evaluation of individual players. Everyone's an All-Star in my eyes at the moment.)